基于LBP纹理分析的视频车辆检测研究

Video Detection Based on LBP Texture Analysis

视频交通事件检测与处理技术在智能交通领域中的应用研究已在学术界和工业界引起了广泛关注。作为视频交通事件检测技术的基础，车辆检测也已成为研究热点。该文主要研究在车辆检测过程中出现的车辆检测不完整，特征信息提取不完全等情况，研究利用LBP算子进行区域生长方法，使得不完整的车辆区域按照LBP纹理方向扩展，从而获得较完整的车辆目标。实验结果表明该算法能够有效进行区域生长，检测车辆的完整度有较大提高。

Video traffic incident detection and processing has attracted wide attention in both academic and industry in the field of intelligent transportation. As the base technology of video detection and processing, vehicle detection has become the research focus. Commonly the vehicles are detected incompletely and thus the features are not extracted exactly. The LBP texture operator is used here to achieve the vehicle region growing to make the incomplete vehicle region expand to the direction of regions with similar LBP texture. Experimental results show that the algorithm can fulfill the vehicle region growing effectively and the integrity of vehicles has improved greatly.